The Decorator's Hand
A deep dive into Shroud of the Avatar's home decorating, what Ultima Online and every housing game since learned about the tools of placement, and twenty-six concrete proposals.
Decorating is the quiet giant of this game. The engine under it is genuinely strong: 6,524 placeable decorations, free move, rotate, scale, and flip, items nested on items eight layers deep, per-item permissions no MMO in our comparison matches, magic movers that carry a whole household between lots, and a grid snap that shipped this month. What the engine never got is the toolkit around the placing itself. A decorator still works one item at a time, with no undo, meters that keep their numbers in tooltips, failures that flash red without saying why, and an edit mode whose keybind ships unbound. This document measures our decorating against the genre that invented player housing and proposes twenty-six changes built on one observation: the engine has been generous for years. It is time the tools caught up.
Ultima Online shipped player housing with the game in 1997, and decorating was literally "drop an item and lock it down." Five years later the single most demanded retrofit in UO's history was the Age of Shadows custom house tool. The twenty-five years since have been a genre-wide arc of the same lesson learned over and over: EverQuest 2's layout files became community infrastructure, Wildstar set the free-placement gold standard, FFXIV shipped a beautiful preview system and then watched its players float-glitch furniture anyway, ESO retrofitted a 200-step undo a year after launch, Palia converted casuals with a build camera, and in late 2025 WoW's housing shipped with an official constraints-off Advanced mode on day one. The verdict is consistent: decorating demand is never satisfied by content; it is satisfied by tools.
Measured against that bar, Shroud's decorating is an inverted picture: the underlying machinery is ahead of most of the comparison column, and the toolkit on top of it is a decade behind. The gaps cluster in three places:
The deal that runs through all twenty-six proposals: tier 1 makes the existing system say what it knows, for almost no engineering cost. Tier 2 adds the verbs every 2026 reference system has, and adds them cheaply because the batching, before-state capture, and permission machinery already exist underneath. Tier 3 finishes the paper features and prices the limit raise in telemetry instead of hope. A decorator who has fought the tool for years should feel the difference in the first evening; a decorator who never started should finally have a reason to.
| # | Proposal | Tier | Size | Server | One-line pitch |
|---|---|---|---|---|---|
| D1 | Placement denials that speak Built | 1 | S | none | surface the failure reasons the engine already computes |
| D2 | Make deco mode findable | 1 | S | none | default keybinds, options rows, and key hints for features players cannot find |
| D3 | A smarter palette Built | 1 | S | none | category filter, recents, click-to-highlight, readable locations |
| D4 | Meters that tell the whole story | 1 | S | none | exact counts and caps on the bars; the community spreadsheet era ends |
| D5 | Who placed this Built | 1 | S | none | owner and last-mover stamps, already stored, finally shown |
| D6 | A front door for deco requests | 1 | S | process | one tracked intake for the busiest request firehose in the game |
| D7 | One tested placement core | 2 | M | none | the enabler: extract the placement math into a pure, tested module |
| D8 | Undo and redo Built | 2 | L | none | a real undo stack over before-states the client already captures |
| D9 | Multi-select and group move | 2 | L | none | marquee selection; the whole dining set moves as one |
| D10 | Copy transform and eyedropper | 2 | M | none | per-channel transform paste, click-to-clone, duplicate-in-row |
| D11 | Magnet snap and align tools | 2 | M | none | item-to-item flush snapping; straight lines without spreadsheets |
| D12 | Precision mode | 2 | M | none | nudge keys, typed transforms, and a parent chain for deep stacks |
| D13 | Ghost preview and dry run | 2 | M | none | know before you commit; the validation primitive layouts need |
| D14 | Free placement mode | 2 | M | none | an advanced toggle that relaxes placement assists; limits stay on |
| D15 | A decorator camera | 2 | M | none | overhead and free-fly views inside deco mode |
| D16 | Decoration layouts and layout slots Built | 2 | L | none | save and re-apply named layouts; the seasonal swap becomes one click |
| D17 | Batch permissions | 2 | M | none | set permissions on 32 chess pieces in one dialog |
| D18 | Finish the flags | 2 | L | data | the 1,750 flip-flagged items and the surfaces that accept nothing |
| D19 | Basement and dungeon pass | 2 | M | data | dungeon wall surfaces, the NPC-allowance bug, sturdier placement zones |
| D20 | Limits that grow with the lot | 3 | M | perf gate | the limit raise as a published, reversible, telemetry-gated schedule |
| D21 | Wider item scaling | 3 | S | perf gate | wider scale ranges on audited families, under the same gate |
| D22 | Move permissions, for real | 3 | M | server | ship the "not implemented" move mode and split use from move |
| D23 | Bigger access lists | 3 | S | server | raise the 10-name cap to 50, with a new server guardrail |
| D24 | Deco NPC patrol routes | 3 | L | saved key | waypoint routes recorded in deco mode; lot NPCs come alive |
| D25 | Shareable layouts Built | 3 | L | item | tradeable layout items; decorators become a profession |
| D26 | Governor terrain leveller Built | 3 | M | data | the deferred placeable that flattens ground for town builders |
The pieces in place are real, and recent work has been steady: optional grid snap shipped in July with four cell sizes and 15 degree rotation steps, elevated placement opened 174 large structures to balconies and platforms, the mirror flip key became a true mirror and 1,250 decorations can use it after an automated audit of all 6,524, and the deco palette moved to the new UI toolkit and gained a search box on the way. Underneath, every placed item remembers its position, rotation, scale, flip, nesting parent, per-item permission groups, owner, and last mover, and every change travels in one atomic batched update.
Here is the entire decorating interface, faithfully. This is what a City lot owner sees after placing their seven hundredth item:
Every decoration you place or move runs the same path, and understanding it is understanding why the proposals in this document are cheap:
flowchart LR A["Palette drag\nor lot click"] --> B["Placement state machine\ncarries ONE item"] B --> C["Per-frame solve\nsurface, snap, input,\nvalidity"] C --> D["Validation\nreasons computed,\nthen hidden"] D --> E["Batched update\natomic, any size,\none server call"] E --> F["Every visitor sees it\nrevision-checked"] style B stroke:#a33d2e,stroke-width:2.5px style E stroke:#c9a55a,stroke-width:2px
The red box is the bottleneck: the state machine was built around a single held item, which is the real reason there is no multi-select, no group move, and no layout apply today. The gold box is the opportunity: the update operation already accepts any number of items in one atomic call, already resolves editing conflicts by revision, and already lets the client capture the previous state of everything it touches. Bulk tools, undo, and layouts need almost no new server machinery; they need the client to stop working one item at a time.
Built recently, credited here, not re-proposed: grid snap (tracker #907), elevated placement on 174 structures, the true mirror flip with 1,250 items enabled, whole-decoration distance culling, five shipped phases of the town performance pass, the mesh-baking system now awaiting its audit, tip jars, tavern game tables, player-named basements and dungeons, the vacate warning with basement item counts, search-all-my-stuff across lot chests, the crafting pull from lot supply chests, eviction of banned players under mutual PvP flags (in progress), the global dev-placed decoration system, terrain flattening under town lots and houses, and the AI dungeon builder's decoration engine. Granular guild property permissions are a separately planned program (#2331) and are credited to it. If you have asked for one of these before: it exists or it is on its way.
UO's housing shipped with the game in 1997, and its decorating system was two verbs: drop an item, lock it down. That was enough to ignite the most durable player behavior in MMO history. Five years of pent-up demand later, Age of Shadows shipped the custom house design tool, a full tile-based editor bolted onto a live game, and it remains the most celebrated retrofit UO ever made. The years after taught the quieter lesson: UO added an interior decorator wand for raise and turn, and players still stacked goza mats into scaffolding to fake vertical placement, then axed the mats. When tools lag demand, players build their own physics exploits. FFXIV learned the same lesson twenty years later when float-glitching became a community norm with its own wiki genre, and WoW closed the loop in 2025 by shipping the "exploit" as an official Advanced mode on day one. Twenty-nine years on, UO's live team still ships decorating features, and its veteran rewards are deco items. Housing outlived every system around it.
Ultima Online: the ancestor and the argument template; lockdowns proved the demand, Age of Shadows proved a live game can retrofit a real editor, and the goza-mat era proved what players do when it does not. EverQuest 2: layout files as community infrastructure; one export command grew a twenty-year toolchain of shared, remixed, and sold house layouts, plus the genre's biggest item budgets. Wildstar: the free-placement gold standard; every item movable and scalable on all axes, numeric coordinate entry, and copy and paste of transforms with per-channel toggles; nobody has matched it since. FFXIV: the preview system done right, including previewing furnishings you do not own; also the cautionary tale, because strict placement rules made float-glitching the community standard for serious builds. ESO: the retrofit masterclass; a year after housing launched it added undo with a visible 200-step history, item linking for group moves, precision numeric mode, and buying furnishings from inside the editor. Rift: drag-handle scaling turned boulders into mountains and plates into floors, and shareable dimension keys built a curation culture that outlived the dev team. Palia: the approachability benchmark; an overhead build camera, one-key grid and freeform toggle, and group select convert non-decorators in minutes. The Sims 4: the non-MMO benchmark every player has used; undo, the eyedropper, style filters, and the official cheat that turns validation off, which is the feature request every housing MMO eventually receives. The meta-pattern across all eight: decorating demand is never satisfied by content; it is satisfied by tools.
| Capability | UO | EQ2 | Wildstar | FFXIV | ESO | Rift | Palia | Sims 4 | SotA today | SotA proposed |
|---|---|---|---|---|---|---|---|---|---|---|
| Undo / redo | ○ | ○ | ○ | ○ | ● 200 steps | ○ | ○ | ● | ● 50 steps, built | ● D8 |
| Multi-select / group move | ○ | ◐ offline tools | ◐ mounting | ○ | ● linking | ○ | ● group select | ● | ○ | ● D9 |
| Copy transform / eyedropper | ○ | ○ | ● per-channel | ○ | ◐ addon | ○ | ○ | ● eyedropper | ○ | ● D10 |
| Decoration layouts / layout sharing | ○ | ● layout files | ○ | ◐ third-party | ◐ addon | ○ | ○ | ● gallery | ○ movers only | ● D16/D25 |
| Align / magnet to neighbor | ○ | ○ | ◐ mounting | ○ | ○ | ○ | ◐ attach points | ● slot snap | ○ | ● D11 |
| Grid + precision snap | ● tile native | ○ | ● numeric | ◐ | ● precision mode | ◐ gizmo | ● grid mode | ● | ● shipped #907 | ● + options UI |
| Constraints-off tier, official | ○ goza era | ◐ blocks | ● by design | ○ float glitch | ◐ | ● scale freedom | ◐ freeform | ● moveobjects | ○ | ● D14 |
| Budget visibility by tier | ◐ house sign | ● | ◐ | ● | ● 4 counters | ● | ◐ | ○ n/a | ◐ tooltips only | ● D4 |
| Placement failure feedback | ◐ generic | ◐ | ◐ | ◐ generic | ◐ generic | ◐ | ◐ | ● reasons | ○ red flash | ● D1 |
| Preview / dry run | ◐ house only | ◐ | ● | ● unowned too | ● in-editor buy | ◐ | ● ghost | ● | ◐ held ghost | ● D13 |
| Palette search and filters | ○ | ◐ | ◐ | ◐ | ● catalog | ◐ | ● categories | ● style filters | ◐ name search | ● D3 |
| Per-item permission granularity | ◐ | ◐ house-wide | ◐ | ◐ house-wide | ◐ house-wide | ◐ house-wide | ◐ shared plot | ○ n/a | ● ahead of genre | ● + batch D17 |
| Co-op decorating | ◐ friends | ● trustees | ● neighbors | ● tenants | ● decorator role | ● edit rights | ● shared plots | ○ n/a | ● roles today | ● + audit D5 |
| Seasonal layout swap | ○ | ● multi-layout | ○ | ◐ third-party | ◐ addon scenes | ○ | ○ | ◐ manual | ○ | ● D16 |
● native · ◐ partial, era-limited, or third-party · ○ absent. Cells reflect each game's official shipping tools as best documented. "SotA today" = the current development branch, 2026-07. Note the two columns where SotA already leads: per-item permissions and co-op roles. The matrix is not a wall of shame; it is three strong columns surrounded by missing verbs.
Tier 1 makes the system say what it already knows: reasons, numbers, names, keybinds. Tier 2 adds the verbs, built on machinery that already batches, already remembers, and already arbitrates. Tier 3 finishes the paper features and prices the limit raise in telemetry. Every tuning number on this page is an illustrative starting point, and one rule holds everywhere: these tools change how you place, never what you may place. Limits, permissions, and ownership checks stay on in every mode this page proposes.
Built 28 July 2026. Every refusal now names its reason on the deco strip and under the held item, the count limits say which one is full and by how much, and the transform keys on multi-part decorations answer instead of doing nothing. Not on the test shard yet. Written up in the work blog for 28 July.
When a decoration refuses to place today, the game says nothing: the ghost stays red and you guess. The engine already computes the exact reason every time (wrong surface, outside the placement area, over a limit, bad nesting parent) and even formats it into a message, but the message is compiled out behind a flag only developers see. This card routes the top reason to a one-line status readout in deco mode: "This item needs a wall." "Outside the basement placement area." "Interior object limit reached: 1000 of 1000." It also ends the silent lockout on multi-part decorations, which today refuse to scale, flip, or roll with zero feedback, so players reasonably assume the keys are broken. Threads like 180508 (the kitchen sink that refuses to exist outdoors, 5 replies) and 182171 (six bowls that accept nothing) are downstream of denials nobody can see.
Seen in: The Sims 4 and WoW Midnight show explicit invalid-placement reasons on the ghost; FFXIV pairs its tint with messages.
The plumbing exists end to end: the validation pass yields structured status entries, an aggregator formats them, and a delivery hook already carries them to the placement context. The work is authoring player-facing text for each reason, routing the top-priority one to a small always-on label on the strip, and adding one honest line at the three multi-part guards: "Multi-part decorations cannot flip, scale, or roll." A test pins every validation branch to a non-empty message so no new rule can ship mute again.
In July 2026 a player publicly asked for a dedicated decorating edit mode, and the honest answer was that it already exists as an unbound keybind (thread 182067, answered within two days). That thread is the whole card in miniature: the deco mode toggle and the palette toggle both ship unbound, grid snap lives on two keys with no options row, and the scale keys appear in no menu, tooltip, or hint anywhere. This card binds the mode and palette by default, adds options rows for grid snap and cell size, puts a compact key-hint line on the deco strip, and makes rotate-while-held honor your rebindings, which it currently does not. Capability without discoverability is capability the players do not have.
Seen in: Palia (one visible keypress toggles build mode), The Sims 4 (every tool shows its key).
| Action | Today | Proposed |
|---|---|---|
| Toggle deco mode | unbound | bound by default, shown on the strip |
| Toggle deco palette | unbound | bound by default, opened by the Place mode button too |
| Grid snap on/off, cell size | G and Alt+G, no options row | same keys, plus options rows and a strip chip showing state |
| Scale up/down | Alt+X and Alt+Z, documented nowhere | shown in the strip key hints and the precision inspector |
| Rotate while held | fixed keys, ignores rebinding | fully rebindable |
Built 29 July 2026. The palette can now find things. Pick a row and press Find, and the decoration is marked on your compass and on the map, pinned to the edge of the strip when it is behind you, and glowing gold by the time you reach it; the marker clears itself when you arrive. Double-clicking a row does the same, and so does Find on the right-click menu. The Location column stopped printing coordinates and now reads "Interior, on a wall, front left", which works even for decorations you have hidden. There is a category filter offering only the kinds your lot actually has, and a Recent row of what you last placed or moved, where clicking a name selects that decoration and points at it. Two pieces of the pitch below are missing on purpose: the game has no idea which floor of a house something is on, and nothing in it marks a decoration as a plant, so the card ships the parts that are true. Find is shown to the same people the Location column is.
The deco palette gained a search box in its move to the new UI toolkit, and that is where modernization stopped: no category filter, no recents, and a Location column that prints raw coordinates like "(12.3, 0.0, -4.1)". This card adds a category dropdown (containers, lights, NPCs, plants, surfaces), a recently-placed rail for the "one more sconce" workflow, click-a-row to highlight the item in the world, and location text a human can read: "Interior, Floor 2" instead of a vector. Notably, zero forum threads ask for any of this, which we read as opportunity rather than absence of demand: nobody asks because nobody expects it.
Seen in: ESO's in-editor catalog browsing; The Sims 4 style and pack filters.
| Item Name | Location | Visibility |
|---|---|---|
| Wall Sconce | Interior, Floor 1 | Visible |
| Wall Sconce | Interior, Floor 2 | Visible |
| Wall Sconce | Exterior | Visible |
| Garden Gnome | Hidden | Hidden |
| Reaper Statue | Exterior | Visible |
The Storage and Layouts tabs and the group-drag footer belong to later cards (D9, D16); the mockup shows the finished palette once the family is complete.
The deco strip shows progress bars, but the community still keeps spreadsheets of the real caps: thread 180910 has players reciting the entire limit table from memory because the game shows it nowhere. This card puts the numbers on the bars themselves ("743 of 1000"), adds the missing dungeon-points meter, colors the bar as you approach a cap, and gives each category a plain-language tooltip explaining what counts against it. It is also the groundwork for the limit conversation in D20: budgets cannot be renegotiated while they are invisible.
Seen in: ESO's four-counter budget display; FFXIV's explicit per-size budgets.
Built 28 July 2026. Hovering a decoration now names who placed it and who moved it last, with the time read back as "3h ago" for anything recent and a date after that. Those two lines existed before but were thrown away whenever the name had not arrived from the server yet, which on a first visit to a lot was every time; the game now asks for the names as your cursor reaches the item, and says what it knows even when a name is still on its way. The decoration list gains a Placed By column, the placement settings window explains what each access level actually allows, and holding Ctrl over a decoration says who its permission tier lets move it. Not on the test shard yet. Written up in the work blog for 28 July.
Every placed decoration already carries its owner and its last mover, stamped and verified by the server, and the client never shows any of it. Town stewards who lost items when another steward redecorated (thread 179995: "I want some security in knowing that some other steward won't come in and clear the items out") and households decorating together get no answer to the simple question "who moved my chair". This card surfaces owner, last mover, and last-moved time in the decoration tooltip and as an optional palette column, and adds a short in-game explainer of what Kindred, Trustee, and per-item permissions actually allow. Granular guild rights are already a planned program (#2331) and are credited to it, not duplicated here.
Seen in: FFXIV shared estates show placer attribution; EverQuest 2 trustee culture runs on it.
New decoration items are the loudest ask in the entire forum: 46 threads in eighteen months plus a twelve-page pinned mega-thread still active this year, and roughly one hundred pure-decoration requests that our own research filters historically skipped precisely because they carry no mechanic. This page is about tools, so it carries exactly one content card: a formal intake. A pinned request board with a visible status column, a recurring batch that converts existing world props into placeable items (the cheapest content in the game; the model already exists), and a public tally so the Elven Mirror or the decoy basement door is a tracked row instead of page twelve of a sticky. The climate already proves conversion works: one multi-ask thread this July had an item shipped the same week.
Seen in: FFXIV furnishing design contests; EverQuest 2 community item batches; UO's veteran-reward deco cadence.
Every placement decision in the game runs through one 490-line method that carries exactly one item at a time and reads the keyboard in the middle of its math. That shape is the real reason multi-select, previews, and free placement are hard today. This is the invisible card that makes the visible ones cheap: extract the movement math, surface classification, and validity checks into a pure, engine-free module with the same treatment the grid snap math already received, including its own test suite. Players never see this card; they see that every card after it ships faster and breaks less.
Seen in: nowhere, and that is the point; this is our own grid snap precedent applied to the rest of the pipeline.
Built 28 July 2026, straight to the full card. Moves, stows, pick-ups, and placements all undo with Ctrl+Z and redo with Ctrl+Y, up to 50 steps. The deco strip gained undo and redo buttons and a list of your recent changes. Undoing a pick-up puts the very same item back exactly where it stood, and if another decorator changes the lot, the history clears with a message instead of guessing. Not on the test shard yet. Written up in the work blog for 28 July.
Every modern decorating system has undo; ESO shipped a 200-step history in 2018, and its players had built an addon for it before that. Shroud has none, and yet the client already remembers the previous state of every move it makes, and the update machinery already accepts a rollback. One misclick on a 300-piece build currently means reconstructing from memory. Version one is single-step: undo the last move, placement, or removal. The full card is a stack of fifty or more steps with a small history panel, where undoing a removal puts the item back exactly where it stood.
Seen in: ESO Update 20 (200 steps, visible history), The Sims 4, WoW Midnight.
The seams are already there: the move flow captures the original settings of every item it touches, and the batched update transaction takes arrays with a rollback delegate. An undo entry is simply the item and its before-state, replayed through the same operation. Removal-undo snapshots the state at pickup, before it is cleared. If someone else edits the plot between your action and your undo, the revision check catches it: the stack refreshes with a status line rather than half-applying, which is the honest behavior on a shared lot.
"If I want to give an occasional concert in my castle courtyard, I have to dismantle the floor and everything on top of it to be able to set up the stage" (thread 182109). Moving a dining set today is eight separate drags. This card adds a Select mode to the deco strip: marquee and shift-click, then move, rotate, or pick up the whole set as one rigid group, committed as one atomic batch. The update machinery already carries any number of items in one call, so the server side of this card is already built; the work is teaching the one-item-at-a-time placement state machine to hold a crowd.
Seen in: ESO item linking (their launch demo was a 30-piece gaming table), Palia group select, The Sims 4.
A new group-move state uses a ghost pivot with children keeping their offsets, validated by the extracted core's batched checks instead of one full pass per item per frame. Selection lives in its own mode so it cannot fight the existing click-to-move. Nested arrangements move with their parent. Two hygiene rules are load-bearing: planters recompute their greenhouse growth bonus after relocation, and the limit counters update as a set so the strip meters never lie. Group actions respect every per-item permission; items you cannot move simply do not join the selection, and the readout says how many were skipped.
Wall-stacking identical shelves today means eyeballing every one. This card adds copy and paste of a decoration's transform with per-channel toggles (position, rotation, scale), an eyedropper that clicks any placed item and readies an identical one from your palette or inventory, and duplicate-in-row for fences, hedges, and pavers placed in runs. It is the tool the block-builder crowd described in thread 182039's replies, where the alignment ask extends to casks, fences, and hedges laid out in lines.
Seen in: Wildstar's copy and paste transform with channel toggles; The Sims 4 eyedropper.
Grid snap shipped in July and is credited, but the hottest placement thread of the month asked for something adjacent: item-to-item snapping. "Placing every block manually can be incredibly tedious. The whole structure ends up looking crooked" (thread 182039). This card adds a magnet latch that snaps a held item flush to the faces and edges of its neighbors, plus align and distribute commands over a selection: align left, center on axis, space evenly. It also promotes the per-surface snap points that already exist in the item data from an invisible accident into a visible choice.
Seen in: WoW Midnight's surface align, Palia attach points, and every 3D tool players compare housing editors to.
For the finishing pass, decorators need millimeters, not mouse drags. This card adds arrow-key nudging in configurable steps and a small transform inspector: typed X, Y, Z position, rotation, and scale, plus a parent-chain readout with select-parent and select-child, because decorations nest eight layers deep and deep stacks are archaeology today. Scale and rotation limits per item stay exactly as they are; the inspector is a finer hand on the same rules, and it is also where multi-part items finally say plainly why scale and flip are unavailable.
Seen in: Wildstar numeric coordinate entry; ESO precision mode.
Placement today commits on release and tells you afterward whether it worked. This card makes the held ghost a true preview: green when every rule passes, red with the D1 reason line when one fails, rotation and scale applied live, and nothing sent anywhere until you commit. It also builds the quiet primitive later cards need: a dry run that validates a whole batch of prospective placements and returns a report, which is exactly what applying a layout has to do before a single item moves.
Seen in: FFXIV's preview system, Palia, WoW Midnight ghost placement.
Every under-tooled housing system grows a physics-exploit culture: UO players stacked goza mats into scaffolding, FFXIV players float-glitch furniture as a community norm, and in late 2025 WoW simply made floating official in its Advanced mode. Shroud can skip the exploit era entirely. This card adds an Advanced Placement toggle that relaxes the placement assistant's surface-matching and overlap rules for decorators who know exactly what they want: floating candles, sunken planters, interlocked geometry. What it deliberately keeps: lot containment, because on open-world lots your build is your neighbor's skyline, and every limit and permission, unchanged. This mode frees geometry, never budgets.
Seen in: WoW Midnight Advanced mode, The Sims 4 moveobjects as the official-cheat precedent, Wildstar free placement by design.
The mode is a flag consulted by the extracted validation core: surface and overlap checks relax; the containment checks for lots, basements, dungeons, and foyers always run; count limits and permission gates are untouched. The storage format already supports every arrangement this mode can produce, which is why the card is client work. The status line names the mode while it is active so screenshots explain themselves.
Fighting a third-person camera is the number one casual decorating pain, and the reason the courtyard-concert player dreads large rearrangements. This card adds two camera modes inside deco mode: an overhead build view for layout work, and a free-fly camera for detail work on upper floors and ceilings, both confined to your lot. Palia proved the overhead camera converts non-decorators into decorators in minutes.
Seen in: Palia overhead placement; The Sims 4 top-down; Rift's dimension camera.
Built 29 July 2026, the full card. The housing strip has a Decoration Layouts button. Save your lot's arrangement under a name, as many as fifty, and each layout is a file in your game data folder you can copy or share. Applying one shows a preview first: how many items will move, how many will be placed from your bags, how many will return to your backpack, and which items cannot be applied and why, using the same denial reasons and count limits live placement uses. The preview says plainly that overlap and surface fit are not checked. The apply itself lands as one undo step, so a single Ctrl+Z walks the whole thing back. Two saved names make the seasonal swap. Not on the test shard yet. Written up in the work blog for 29 July.
EverQuest 2 shipped a save-layout command two decades ago and an entire community toolchain grew on top of it; FFXIV and ESO players had to build third-party equivalents. Our magic movers carry a whole household between lots but cannot save an arrangement: restoring evicts what is there now, and there is no preview. This card adds true layout capture: save your lot's arrangement to named slots, then re-apply any slot through a reconciliation report that lists what matches, what is missing, and what will move where, shown as a D13 dry run before a single item shifts. Named slots make the seasonal swap real: keep "Everyday" and "Yule" and switch in one sitting instead of two weekends. Layouts place only items you own; nothing is duplicated, and the magic movers keep their job.
Seen in: EQ2 layout files and multi-layout saves; ESO's scene addon proving unmet official demand; FFXIV's third-party layout tools.
| Layout | Items | Owned | |
|---|---|---|---|
| Yule Great Hall | 220 | 214 of 220 | Preview Apply |
| Harvest Market Stalls | 64 | 64 of 64 | Preview Apply |
Capture reads the lot's placed items into a local file; the stored per-item record practically is a layout format already. Apply matches by exact item first and item type second, then commits in measured chunks through the ordinary update and transfer paths, with planter growth bonuses recomputed and one compound undo entry pushed for the whole apply. Layout limits are validated in the dry run before anything moves. House-model mismatches reuse the magic mover's existing warning pattern. Local files only in this card; trading layouts is D25's business.
The chess-set problem: a player who places a 32-piece set and wants visitors to move pieces but not pocket them currently opens 32 permission dialogs (thread 178664's ask, echoed in the wishlist survey). With multi-select in hand, this card applies permission changes across a selection in one dialog and one atomic operation: move rights, remove rights, container put and take. The specific "public move, not take" setup works end to end today; it is purely unreachable through the one-at-a-time interface.
Seen in: nowhere good, honestly; bulk furnishing permissions are a genre-wide gap Shroud can leapfrog because its per-item model is already deeper than the norm.
Mirror flip shipped this month with 1,250 items enabled after an automated audit of all 6,524 decorations, and players noticed within days. The same audit flagged about 1,750 items that need per-item fixes before they can mirror safely. Meanwhile a parallel coverage gap sits in surfaces: six named bowls that accept nothing ("It would be nice to be able to put things inside them", thread 182171), garden bowers that should take wall decorations, an energy capacitor stuck floor-only, a kitchen sink that refuses to exist outdoors. This card is the workstream that finishes both: scheduled remediation batches through the audit tooling that already exists, with a public running count so progress is visible patch by patch.
Seen in: nothing to transplant; this is finishing our own shipped systems, and crediting the audit that already paid for the map.
Player dungeons and basements are where the most ambitious builds live, and where the sharpest edges are. The Stone Dungeon Rotunda cannot take wall lights or chandeliers (thread 179899, 2,429 views, acknowledged on the team's tracker and worked on-stream), the Duke Throneroom's surfaces leave almost no safe space to build on, and this review found a genuine bug: dungeon NPC allowances are computed from the lights value instead of the NPC value. The pass: add wall and ceiling surfaces to the acknowledged rooms, fix the allowance bug, and harden the placement-zone lookups underground so they fail loudly instead of quietly.
Seen in: closing our own tickets, with the community's numbers attached.
Two years ago a request to scale outdoor decoration limits with lot size was denied for two stated reasons: the rendering cost of lights, and the loading stall on decoration-heavy town lots. Both were true, and the team has spent 2026 dismantling them: whole-decoration distance culling is live in four bands, five of seven town performance phases have shipped (including an instancing pass that cut draw calls on dense agriculture fields by 94 percent), and the mesh-baking system that merges repeated outdoor decorations is code complete and awaiting its audit. This card proposes the raise as a schedule, not a wish: when the baking system exits testing and the measurements stay flat on the heaviest audited lots, limits step up on a published table, and each step is reversible if the telemetry regresses. The gate criteria go on this page so players can watch the same numbers the team does.
Seen in: ESO and FFXIV both moved caps only alongside performance work; publishing the gate is our own addition.
| Lot tier | Ext objects today | Step 1 | Ext lights today | Step 1 |
|---|---|---|---|---|
| Village | 400 | 500 | 6 | 9 |
| Town | 700 | 875 | 8 | 12 |
| City | 1000 | 1250 | 12 | 18 |
| Keep | 1400 | 1750 | 16 | 24 |
| Castle | 2000 | 2500 | 20 | 30 |
Illustrative step 1 numbers: plus 25 percent exterior objects, plus 50 percent exterior lights, Village and above. Gate: the mesh-baking audit ships clean, scene load times stay flat on the ten heaviest audited lots, and frame captures on those lots hold their current budget. A second step waits on the remaining performance phases. Every limit is a data value, so each step is a small, reversible change.
Rift turned a modest catalog into infinite building material with one mechanic: scale anything, tiny to giant. Shroud already scales items, but the per-item ranges are conservative. Under the same performance gate as D20, this card widens scale ranges on audited item families: statues, rocks, plants, architectural pieces. A boulder at three times scale is a hill; a plate at five times is a table. Scale multiplies rendering cost, which is why this card shares D20's gate instead of pretending otherwise.
Seen in: Rift's drag-handle scaling, the beloved one; Wildstar; WoW Midnight Advanced.
The permission matrix has two long-standing paper features. The lot-wide "Public" move mode has been marked "not implemented" in the player documentation for years, and the per-item move permission is locked to one hardcoded tier, with the option that should control it still a comment in the code. The result is real grief: "There is no way to use planting beds without the right to pick them up" (thread 179650, confirmed by a player who tested every permission combination). This card finishes the matrix: implement the Public move mode, give the per-item move permission real tiers on both client and server, and separate use rights from move rights so a guest can tend a planting bed without being able to walk off with it.
Seen in: FFXIV estate permission tiers; our own design docs, which have carried the "not implemented" note for years.
The bones already exist on both sides: the mode value, its wire key, its database field, and the operation that sets it all ship today; the move path just never consults them. The work is consulting them, honoring per-item tiers in the same check, replacing the client hardcode with the option it always wanted, and adding the third state to the lot permissions window. Use-versus-move is client interaction gating decoupled from the move check. This is the smallest server card in the program and it retires the oldest documented gap.
"I have found I am limited to a tiny 10 people on access lists to my properties" (thread 180850, 16 replies, the hottest permission thread of the winter). Doors and chests cap at ten names, vendors at thirty, and one of those caps is annotated "arbitrary limit" in the source. This card raises the caps to fifty across the board and pairs the raise with a new server-side length guardrail, so bigger lists arrive tidy end to end rather than merely permitted.
Seen in: ESO access and ban lists run to hundreds; FFXIV estate access similar.
| Surface | Today | Proposed |
|---|---|---|
| Doors | 10 | 50 |
| Chests and containers | 10 | 50 |
| Player vendors | 30 | 50 |
Eight forum threads want lot NPCs that feel alive instead of statuary: conversationalists that sit, walk a route, or at minimum stop clipping through the furniture. The drift bug where they wandered off their spot was fixed this month and is credited; routes were never built. This card adds a route recorder in deco mode: select a decoration NPC, drop numbered waypoints, set pause times, done. Routes save inside the decoration's own record, so they persist, move with magic movers, and need no new server systems. Walkers respect the same furniture obstacles the pathing system already carves, which quietly answers the clipping complaints too.
Seen in: The Sims 4 NPC schedules; Wildstar housing actor paths; EQ2 house actors.
Built 29 July 2026, together with D16. The Decoration Layout is a craftable item: Tailoring skill 30, twelve Wood Pulp and four Cotton, with the recipe sold by the merchants that carry the blank book recipe. Use a blank one on a lot you may decorate to record the layout under a title, and the written layout shows that title as its name, like a player-written book. It trades, mails, and sells on vendors like any item. Whoever holds it can preview and apply the arrangement to a lot they may decorate, using their own decorations. A layout carries the arrangement, never the objects, and a written one cannot be recorded again. Not on the test shard yet. Written up in the work blog for 29 July.
EverQuest 2's layout files became community infrastructure: shared, remixed, and sold for twenty years. Once D16 gives Shroud layout capture, sharing is the natural second act: a craftable Decoration Layout item that stores a captured layout and can be traded, mailed, or sold on a vendor. The buyer applies it through the same reconciliation dry run, using their own items; a layout carries arrangement, never objects. A furniture economy grows a design economy on top, and decorating becomes a profession with a signature.
Seen in: EQ2's layout-file economy; FFXIV's third-party layout ecosystem proving the demand; Rift dimension keys.
Built 28 July 2026. Three sizes of Terrain Leveller, 8x8, 16x16 and 32x32 meters. A governor or steward drops one and the ground across its square flattens to the height of the marker, with the grass and trees inside the square cleared. Each counts as one town object, and towns get a leveller cap that scales with their size. Ground that rises and falls too much across the whole square is still refused, and the message now gives you the measured slope and the limit. Not on the test shard yet. Written up in the work blog for 28 July.
Terrain flattening under town lots and houses shipped this month, and its plan explicitly deferred the follow-up town builders asked for within days (thread 182123): a governor-placeable leveller. This card picks it up: a placeable object for town governors that flattens its footprint, so plazas, markets, and event grounds stop being terrain-request tickets. The flattening capability already exists as a decoration behavior; this gives governors a handle on it, counted against the existing town object budget.
Seen in: Palia's landscaping tools; Rift's terrain-as-decoration culture.
Every approved card composes into one surface. No new floating windows: the strip you already know gains a mode row and its numbers, the palette gains tabs, and a small inspector appears on demand. This is the same bar from section 1, after the program ships.
| Phase | Ships | Rule |
|---|---|---|
| P1 · Speak, find, count | D1, D2, D3, D4, D5; D6 opens its board | Pure client quick wins riding the in-flight strip conversion; immediate goodwill, no dependencies. |
| P2 · Foundation | D7, then D8 (single-step first, stack second), D12, D13, D19; D18 batches begin and never stop | Extraction first; then the cards that need only single-item machinery. |
| P3 · The group era | D9, D10, D11, D17, D14, D15 | Everything selection-shaped lands after undo exists, because group tools multiply the cost of mistakes. |
| P4 · Layouts and the matrix | D16, D22, D23, D24, D26 | Layouts compose the earlier primitives; the two permission cards ship as one server release. |
| P5 · The gated raises and sharing | D20, D21, D25 | Limits and scale step when the published gate clears; shareable layouts after local ones prove out. |
flowchart LR D7["D7 placement core"] --> D9["D9 multi-select"] D7 --> D12["D12 precision"] D7 --> D13["D13 dry run"] D7 --> D14["D14 free place"] D1["D1 reasons"] --> D13 D1 --> D14 D8["D8 undo"] --> D9 D8 --> D16["D16 decoration layouts"] D9 --> D10["D10 copy"] D9 --> D11["D11 align"] D9 --> D17["D17 batch perms"] D9 --> D16 D13 --> D16 D16 --> D25["D25 shareable"] D4["D4 meters"] --> D20["D20 limit schedule"] PERF["perf gate: bake audit + telemetry"] --> D20 PERF --> D21["D21 scaling"] style D7 stroke:#c9a55a,stroke-width:2px style D8 stroke:#c9a55a,stroke-width:2px style D20 stroke:#a33d2e,stroke-width:2px
This page deliberately builds on what shipped recently rather than re-asking for it: grid snap with four cell sizes, elevated placement on 174 structures, the true mirror flip and its 6,524-item audit, whole-decoration distance culling, the town performance pass, player-named basements and dungeons, the vacate warning, search across lot chests, the crafting pull from supply chests, tavern game tables, tip jars, and terrain flattening under lots and houses. If you have asked for one of those before: it exists, go enjoy it.
| Claim | Source surface |
|---|---|
| 6,524 decorations audited; 1,250 flippable; ~1,750 flagged | The flip audit tool's classification report, July 2026 |
| Per-tier limit tables (Village 400/6 through Castle 2,000/20 exterior) | The plot limit data tables and the housing catalog docs |
| The client keeps the before-state of every move | The placement controller's move flow |
| Updates travel as one atomic batch of any size | The placed-item update operation and its transaction |
| Failure reasons computed, then hidden | The placement validation messages behind a developer flag |
| Dungeon NPC allowance uses the lights value | The dungeon limit calculation; flagged as a found bug (D19) |
| Deco mode and palette keybinds ship unbound | The default keybind table |
| Forum demand figures and quotes | The official wishlist and housing forums, threads cited by id |
| Performance receipts (culling bands, instancing gains) | The 2026 performance program's plans and captures |
The grid snap module set the pattern: pure math with its own EditMode suite plus a binding ratchet that pins keys and verifies consumption. Each shipping card extends it: the placement core's extracted math and validation predicates (D7), undo stack semantics and inverse construction (D8), group pivot math and the planter recompute (D9), magnet and align math (D11), the layout reconciliation matcher (D16), and headless UI tests for the strip and meters in the pattern the deco panel tests already use.
Shipping game data and client code (the placement controller, the palette, the limit tables, the flip audit); the housing documentation suite and catalog; roughly 90 decoration threads catalogued from the official wishlist and housing forums for this review, out of about 300 wishlist threads posted since January 2025, with engagement figures per thread; the community wishlist survey; and the genre record: UO's publish history and housing wiki, EverQuest 2's layout-file documentation, Wildstar housing guides, FFXIV's housing documentation and its community glitch guides, ESO's Update 20 notes and UESP housing pages, Rift dimension guides, Palia patch notes, The Sims 4 build documentation, and the WoW Midnight housing coverage.
No account needed. Every proposal above also has its own Discussion box for talking about that one alone; this thread is for the plan as a whole. Comments are plain text, held to basic decency, and may be trimmed.